Table 4.

Top 20 dependency distances comparison between human-written and ChatGPT-generated text in the medical abstract and radiology report data sets.

Category	Medical abstract data set		Radiology report data set
	Human (words)	ChatGPT (words)	Human (words)	ChatGPT (words)
Adjectival modifier	1.5	1.4	1.7	1.6
Case marking	2.2	2.2	2.5	2.4
Nominal modifier	4.2	4.1	4.2	4.0
Punctuation	8.5	8.7	5.6	5.5
Determiner	1.8	1.7	2.1	2.0
Compound	1.3	1.2	1.5	1.4
Root of the sentence	7.3	5.9	3.6	4.0
Conjunct	5.9	4.7	4.5	3.6
Nominal subject	3.9	3.0	3.2	2.8
Coordination	3.7	2.9	2.4	1.8
Direct object	2.5	2.4	2.5	2.6
Numeric modifier	1.3	1.2	1.4	1.3
Adverbial modifier	2.2	2.8	1.7	2.1
Passive auxiliary	1.2	1.1	1.2	1.1
Marker	3.5	2.4	—a	—
Dependent	4.8	4.7	3.7	3.6
Copula	2.0	2.4	1.7	1.6
Clausal modifier of noun	2.3	2.5	2.3	2.4
Auxiliary	1.9	1.7	—	—
Passive nominal subject	6.1	5.2	3.8	3.8
Negation modifier	—	—	1.7	1.8
Expletive	—	—	1.3	1.1

^aNot in the top 20 dependency distances.